Improved Neural Machine Translation with a Syntax-Aware Encoder and Decoder
arXiv:1707.05436
Abstract
Most neural machine translation (NMT) models are based on the sequential encoder-decoder framework, which makes no use of syntactic information. In this paper, we improve this model by explicitly incorporating source-side syntactic trees. More specifically, we propose (1) a bidirectional tree encoder which learns both sequential and tree structured representations; (2) a tree-coverage model that lets the attention depend on the source-side syntax. Experiments on Chinese-English translation demonstrate that our proposed models outperform the sequential attentional model as well as a stronger baseline with a bottom-up tree encoder and word coverage.
Accepted as a long paper by ACL 2017
References in corpus (3)
Cited by in corpus (17)
- Glyce: Glyph-vectors for Chinese Character Representations
- Transforming Question Answering Datasets Into Natural Language Inference Datasets
- CODIT: Code Editing with Tree-Based Neural Models
- Neural Networks for Cross-lingual Negation Scope Detection
- Variational Recurrent Neural Machine Translation
- Improving Aspect Term Extraction with Bidirectional Dependency Tree Representation
- Towards Open Intent Discovery for Conversational Text
- Recurrent Graph Syntax Encoder for Neural Machine Translation
- Learning to Discriminate Noises for Incorporating External Information in Neural Machine Translation
- Neural Machine Translation with Source-Side Latent Graph Parsing
- Evaluating the Utility of Hand-crafted Features in Sequence Labelling
- Neural Machine Translation: A Review and Survey
- Syntax-Enhanced Neural Machine Translation with Syntax-Aware Word Representations
- Source Dependency-Aware Transformer with Supervised Self-Attention
- On Tree-Based Neural Sentence Modeling
- Exploiting Rich Syntax for Better Knowledge Base Question Answering
- A survey of cross-lingual features for zero-shot cross-lingual semantic parsing